Intent Modelling Framework
This research explores how human intent can be inferred and represented using a combination of multimodal neural signals and AI models.
Applied Research in
Neuroscience informed AI
Cogitos Labs is an early-stage applied research lab working at the intersection of Neuroscience, AI and Autonomous Systems.
Our current research work is focused on effective modelling of individual intent leveraging brain computer interface technologies and integrating with safe and reliable autonomous agents
Current Research
We are currently exploring early-stage frameworks, methods and tools for intent modelling in individuals using brain computer interface technologies.
Alignment Assurance Framework
We are investigating methods to evaluate and manage alignment risks in autonomous agentic systems, with a focus on long-horizon behavior, tool use, and human-in-the-loop workflows.
Applications
Our research informs a set of potential applications that we are iteratively exploring and validating.
Neural Intent Decoder
Opensource Model that reads multimodal neural signals and translate them into actionable intents. These can be subsequently integrated into autonomous software of robotic agents for safe and reliable actions
Human-in-the-Loop Feedback Pipelines
Methods and tools for integrating structured feedback into agent workflows to support oversight, correction, and learning.
Alignment Health Assessment
An assessment framework aimed at understanding how an autonomous agent’s observed behavior aligns with stated objectives and individual intent under different conditions.
Work With Us
Collaborate With Us
We are open to research collaborations, early exploratory engagements, and technical discussions with teams working on autonomous agentic systems.
Please note at this stage, we are selective and primarily focused on research-oriented or exploratory collaborations.
Join Us
We’re looking for researchers and engineers motivated to explore the foundations
of intent alignment—especially those with backgrounds in:
– ML alignment, evaluation, or interpretability
– Agent systems and tool-use architectures
– Neuroscience, signal processing, or BCI
Engagements may be full-time or part-time roles depending on fit and availability. All roles are strictly remote.
How We Approach Collaboration
- Scoping – Initial discussions to understand the system or research question — whether related to agents, intent representation, or human–machine interaction — followed by appropriate confidentiality and consent arrangements.
- Exploration & Analysis – Collaborative exploration that may include system instrumentation, behavioral analysis, human feedback studies, or limited neural signal experiments, depending on the domain and research objectives.
- Synthesis & Discussion – Shared analysis of observations and findings, typically documented in a research summary and discussed collaboratively to inform potential next research directions.